In the weeks since the explosive revelation that top U.S. officials inadvertently shared attack plans in Yemen with a journalist on a Signal group chat, fresh questions about the Trump administration’s lax approach to digital security have continued to emerge. On April 20, The New York Times reported that the security breach is even worse than initially understood: Defense Secretary Pete Hegseth had also shared many of the same details about the imminent U.S. bombing strike in Yemen in a second group chat with several family members, a personal lawyer, and others, using his private phone.
The fiasco now known as SignalGate raises many urgent issues related to national security. Communicating classified information via nonapproved channels potentially violates the U.S. Espionage Act, setting messages to automatically disappear contravenes U.S. federal laws on preservation of official records, and officials’ family members and journalists should certainly not be privy to this kind of information. These are huge lapses. But by focusing on National Security Adviser Mike Waltz’s unwitting inclusion of The Atlantic’s editor in chief, Jeffrey Goldberg, in the first chat group, much of the debate has downplayed an even larger problem: the very real possibility that a foreign government or other hostile power was snooping on the devices through which those communications were taking place.
Convened by Waltz, the first Signal group included not only the defense secretary but also CIA Director John Ratcliffe, Director of National Intelligence Tulsi Gabbard, Vice President JD Vance, White House adviser Stephen Miller, and Special Envoy to the Middle East Steve Witkoff, among others. Every one of these figures is an extremely high-value target for foreign espionage. Moreover, at least some of the participants were communicating on personal electronic devices (as was the defense secretary, and at least some of the participants, in his second group chat), and several were traveling overseas at the time. Witkoff, for instance, was in Moscow (although he denies using a personal device during that leg of the trip), while Gabbard was in Southeast Asia. It is highly likely that one or more of the participants’ devices was already being targeted by a U.S. adversary armed with advanced spyware capabilities. And the compromise of just one phone is all it takes. One weak link and all of the chat groups’ discussions could be exposed.
These episodes point to a looming digital security crisis. Even as high-level officials, for convenience or other reasons, rely more on personal devices for sensitive communications, those devices have become increasingly vulnerable to targeted exploitation and surveillance. Once a device is compromised, hostile regimes and other malicious actors can spy on communications regardless of whether the information in question is encrypted. Thankfully, there are steps that the United States and other governments can take to mitigate this risk. But to do so, they must first recognize the real problem with SignalGate: how exposed sensitive communications and private information have become for anyone reliant on a modern smartphone.
MISSED SIGNALS
It is important to clarify that the security failures at the heart of SignalGate were not related to potential weak points in Signal itself. Signal is the gold standard of secure, end-to-end encrypted messaging and employs the highest standard of cryptography. Its codebase is open source and widely peer reviewed. For all these reasons, many government agencies, such the U.S. Cybersecurity and Infrastructure Security Agency, now recommend using Signal for nonclassified communications.
But that does not mean that Signal or any other app is appropriate for discussing highly sensitive plans about imminent military operations on regular, unsecured personal devices. Governments, including the United States, have developed very strict procedures for handling classified information. They require personnel to use tightly managed “work” devices to which access is strictly limited, even within the national security hierarchy. Many governments also create protocols to ensure that officials’ discussion of classified information is restricted to sensitive compartmented information facilities, known as SCIFs, which are structurally designed to prevent outside eavesdropping, guarded by sentries and other access controls, whether inside an agency’s headquarters, in mobile situations (such as in vehicles or aircraft), or in foreign embassies.
By using Signal to discuss U.S. plans for bombing Yemen, Waltz and the other 18 members of the “Houthi PC Small Group” were effectively trying to create a do-it-yourself SCIF. According to The New York Times, the information that Hegseth shared on the two group chats, including “the flight schedules for the F/A-18 Hornets targeting the Houthis in Yemen,” originated from a secure government system at the U.S. military’s Central Command, a system designed for handling classified information. As is now clear, sharing this information proved far from secure. Many of the participants presumably connected to Signal on their personal devices, as evidenced by the fact that Waltz had a journalist in his contact list and Hegseth set up his group, which included his brother, his lawyer, and his wife, before being appointed Pentagon chief. Given strict access controls that are typically implemented on devices cleared for classified communications, it is likely that many of the other senior government officials who were members of these groups were using their personal phones to connect, as well. In doing so, the Signal group members exposed the Pentagon, and the United States itself, to a slew of sophisticated digital security risks.
ENCRYPTION FRICTION
Ever since the advent of the smartphone, governments have developed capabilities to surreptitiously hack into those devices. As I described in a 2022 Foreign Affairs essay, “The Autocrat in Your iPhone,” for many intelligence services, smartphones are irresistible targets: not only do users tend to organize their communications, contacts, financial activities, and general digital presence around their personal phones, they tend to carry them wherever they go. Moreover, efforts to make messaging apps more secure have paradoxically fueled even more invasive forms of surveillance.
Traditionally, governments sought to eavesdrop on individuals by gathering data directly from the telecommunications or Internet service providers’ networks through which their communications passed. But beginning around 2013, following the Edward Snowden disclosures about U.S. government surveillance programs, robust encryption protocols began to spread to ordinary consumer communications, and users began to adopt secure applications such as Signal. That rendered the old approach less productive, pushing intelligence and law enforcement to seek ways to get inside devices themselves. Rather than try to crack advanced encryption protocols, why not find access to the unencrypted end of an end-to-end encrypted communication? To acquire the necessary technology, they turned to a growing number of private mercenary firms, such as NSO Group, a notorious Israeli firm that has over the past decade sold its spyware technology to governments around the world. Mercenary spyware firms employ highly skilled engineers to either develop in-house or purchase software vulnerabilities from gray market firms, known as exploits or zero days, of which manufacturers of consumer devices and applications are not yet aware. They or their clients then use those exploits to hack into the target’s personal phones.
The most advanced phone-hacking technology sold by mercenary firms requires no target interaction and leaves no indication that the device has been tampered with. The technology does not require targets to click on a link or download an attachment in order to trigger an infection. Instead, it takes advantage of flaws in hardware, operating systems, or apps to gain silent access to the inner workings of a device, allowing attackers to then implant their spyware surreptitiously. If the attack is successful, targets are unaware that anything about their device has been compromised and simply continue using it, but now with someone peering over their shoulder.
At their most sophisticated, tools such as NSO’s Pegasus can then offer a “god’s-eye view” of a target’s entire life. Once inside a device, a remote operator can turn on the camera, activate the microphone, and track the location of the device’s owner. The operator can look back in time, too, scrolling through camera rolls or observing previous communications to gather information on a target’s personal relationships and past activities. Most important, with this surreptitious access to the device itself, they can read messages that are end-to-end encrypted. When spyware firms boast that they can “crack” Signal, what they really mean is that they can compromise a device on which Signal is loaded.
Consider the case of Jamal Khashoggi, the Saudi journalist and Washington Post columnist who was executed in a Saudi consulate in Turkey in 2018. Before his execution, Khashoggi had been communicating for months with the exiled Saudi activist and Canadian permanent resident Omar Abdulaziz over what they assumed was a secure, end-to-end encrypted app. But research by the Citizen Lab, the research team I direct that specializes in investigating digital threats, revealed that Abdulaziz’s phone had been hacked with Pegasus, rendering both sides of the conversation completely transparent to Saudi operatives.
This mercenary spyware industry is almost entirely unregulated, and over the last decade and a half, security agencies around the world have used its wares to go on an unbridled hacking spree. Along with other groups, Citizen Lab has shown how spyware such as Pegasus has been used by governments in multiple countries to hack into the devices of political opposition members, activists, investigative journalists, and government officials themselves. Although authoritarian regimes are the most prolific abusers of such technology, our research has uncovered domestic espionage and abuse of spyware in the West, as well, including by the governments of Greece, Hungary, Italy, Mexico, Poland, and Spain.
In principle, senior U.S. government officials’ work devices are managed in accordance with NSA standards and may include NSA-developed technology that is resistant to these infections. But privacy regulations justifiably forbid the NSA from checking the personal devices of government officials—a gap in defenses obvious to any adversary armed with mercenary spyware and contemplating where to aim their precious cyber-espionage assets. Rather than try to break into an NSA-secured device, an undertaking that would be technically complex and extremely expensive, they can simply aim at the target’s personal phone. Why bother clambering over a wall topped with barbed wire when you can just stroll around it and enter through an unlocked gate? Thus, the personal device of any high-level official presents a juicy opportunity for intelligence gathering—bringing into relief the dangerous risks posed by the Houthi PC Small Group chat.
PRISONERS OF THEIR OWN DEVICES
The risks posed by using personal devices extend well beyond the threat of spyware. Most personal smartphones are loaded with apps whose underlying purpose is to harvest information about their users. Every digital scrap is sucked into a huge data vacuum: a user’s location information, WiFi networks, SIM card, and other device identifiers are routinely collected by apps and browsers and fed into a vast data bazaar. Primarily, this data is used for marketing purposes—sold to “big data” firms that target consumers with advertisements. When consumers receive advertisements for a vacation spot shortly after they were casually searching for one online, that’s today’s personal data surveillance economy at work.
But these properties also lend themselves to a more sinister application. Another mercenary surveillance industry has emerged alongside spyware, known as “advertising intelligence” or ADINT, that packages this type of advertising data on potential targets into intelligence products to help governments spy. One such firm, the U.S.-based Fog Data Science, brags in its marketing materials that it can collect “15 billion location signals each day” from “250 million devices” and “tens of thousands” of mobile apps. It touts that it can provide security agencies with the ability to see all the identifiers associated with every device active in a given location in a given time frame.
It is highly likely one of the participants’ phones was already targeted with spyware.
Although the scale of the industry is unclear because of secrecy, it is clear from their own advertising that ADINT firms can gather sweeping intelligence on particular targets of interest. Fog Data Science claims that it can track a specific device’s location history over months or even years, analyzing patterns to identify what the firm creepily calls a target’s “bed-downs.” By fusing data collected from data brokerage firms with other open-source information—including from data breaches circulating on the dark web—these advertising intelligence companies can produce extremely precise dossiers on a person’s primary residences, places of employment, known associates and relatives, medical or therapy appointments, religious affiliations, hotel rooms frequented, travel itineraries, and much more.
This kind of information can be supplied to some of the world’s most notorious despots, hostile foreign adversaries, or even organized criminal groups with the click of a purchase order. Nothing is currently stopping Iran or China from using front companies to purchase information on the detailed movements and smartphone configurations of U.S. officials or suspected intelligence officers and their families and then acting on that information as they choose.
Experiments undertaken by researchers and journalists have shown how access to these real-time data brokerages can reveal highly sensitive information about high-value government targets, including active U.S. military personnel. To get a sense of just how detailed this information can be, consider how much information journalists could uncover in the open domain about members of the Houthi PC Small Group. Investigations by Wired showed that several members, including Hegseth, Waltz, and Chief of Staff Susie Wiles, had left information about their Venmo contacts and transactions either entirely or partially public. Journalists from Der Spiegel used commercially available people-search engines and information from data breaches to reveal passwords, phone numbers, and other information linked to Dropbox, Microsoft Teams, Signal, and WhatsApp accounts owned by Gabbard, Hegseth, and Waltz.
ADINT firms can combine such openly available information with advertising data to present a detailed file on almost any target. In the case of the Signal group chats, hostile governments could have used ADINT services to map any of its participants’ locations and precise movements and then put them under physical surveillance, perhaps to gain covert access to their phones when an opportunity arises. Or such services could be used to undertake reconnaissance on the make and model of their personal devices, or those of their acquaintances, in order to target those devices for hacking using precision spyware exploits. Such detailed files could include known associates of principal targets and contain highly specific information about their habits, devices, and applications, broadening the number of potential clandestine entry points into a sensitive conversation. (Hegseth’s Signal group was reportedly created before his appointment as secretary of defense and at one point included a dozen people “from his personal and professional inner circle.”)
UNSAFE TRAVELS
One of the greatest security risks posed by personal devices occurs in a situation when they are most often used: during travel. To connect to the telecommunications network, a cellphone must first communicate with a nearby cellular tower using its built-in radio. Information from the phone’s SIM card is sent to the network for authentication. Behind the scenes, a wide variety of systems—base stations, mobile switching centers, and other infrastructure—route calls and data to and from the device. Although a plethora of fine-grained information gathered from domestic cellular networks is routinely accessed by government agencies, it is when a mobile user travels abroad that their devices may be particularly vulnerable.
When subscribers roam on another provider’s network, telecommunications companies exchange massive amounts of data every second. Telecom companies can use these signals to retrieve detailed personal information about roaming users, including when a phone number is active on a cellular network, the services they draw on, and their precise location. Access to this international signaling network was once confined to a relatively small club of mostly Western telecommunications firms, and by extension, the state intelligence agencies that colluded with those firms, either informally or through lawful access requests. But a growing number of mercenary surveillance firms have managed to acquire licenses, called global titles, or have found ways to cooperate with existing cellular service providers to enter that club. Once inside, these firms can find someone’s exact location, block or slow down their cellular services, or eavesdrop on their calls and texts. They may even be able to intercept two-factor verification codes to hack accounts or surreptitiously transmit spyware.
Many firms now market this kind of surveillance to a growing number of government agencies. A 2020 Citizen Lab report revealed that Circles, a firm that sells signaling surveillance services, had government clients in Botswana, Chile, Ecuador, El Salvador, Equatorial Guinea, Guatemala, Honduras, Indonesia, Kenya, Malaysia, Mexico, Morocco, Nigeria, Peru, Serbia, Thailand, the United Arab Emirates, Vietnam, Zambia, and Zimbabwe. Another Citizen Lab investigation revealed that between November 2019 and March 2020, millions of malicious signaling requests were sent per month from networks in Saudi Arabia to geolocate the phones of Saudi users when they traveled to the United States.
As noted, several members of the Houthi Signal group were outside the United States while participating in the group’s chats. Secretary of State Marco Rubio was in Canada; Witkoff was in Russia and Azerbaijan during part of the group’s discussions. (Witkoff denies communicating on his personal device on those trips, although he has not said whether he was carrying his personal device and whether it was turned on.) Gabbard was on a multicountry trip across Asia, including Thailand. The Royal Thai Army was once a Circles client, and a 2022 Citizen Lab investigation revealed an extensive government espionage campaign targeting Thai pro-democracy protesters and activists, at least 30 of whom had their devices infected with NSO Group’s Pegasus spyware. It would have been trivial for government clients to acquire the Signal group members’ phone numbers or other identifiers and then use malicious roaming messages to covertly track their locations in real time, disrupt their access to messaging platforms, hijack two-factor authentication checks, or inject spyware into their devices should those targets have opened a poorly encrypted website.
CONTROL OR BE CONTROLLED
Threats to the security of personal devices are made worse by the lack of effective regulation of surveillance vendors. Many of these firms employ the same type of obfuscation techniques as oligarchs and money launderers, making regulation challenging. The result is that dozens of governments and even nonstate actors now possess advanced digital surveillance capabilities that were once limited to only a few great powers and can deploy them largely in the absence of restraints. Were any of the Houthi Signal group members compromised at the time they discussed the Yemen attack plans? Frankly, it would be shocking if they were not.
One can only hope that somewhere, professionals are ensuring that the devices of top U.S. officials are free from mercenary spyware and not being tracked. For the moment, the signs are not encouraging. Instead of launching a full investigation of the security breakdown and holding those responsible to account, the Trump administration has defended the defense secretary’s use of the chat groups and continued to downplay their significance. At the same time, the administration has taken steps to weaken or dismantle government offices designed to protect the United States from foreign digital interference and cyberwarfare, shutting down the FBI’s foreign influence task force and radically downsizing the Cybersecurity and Infrastructure Security Agency, including those responsible for “red teaming” government systems. And in early April, apparently in response to a demand by the right-wing conspiracy theorist Laura Loomer, the president fired the head of the National Security Agency and U.S. Cyber Command—the top U.S. official managing the country’s cyberdefenses.
The spyware industry is almost entirely unregulated.
The consequences of this disregard for device security could soon come back to haunt the administration. To be fair, there are no easy fixes. It would be impractical, and possibly illegal in some jurisdictions, to ban government personnel from using personal devices on their own time. And the only way to truly insulate government officials from the growing number of surveillance risks would be for them to completely avoid using personal devices while they travel—another impracticality. Most government officials have found it extremely challenging to properly do their business without using personal devices in one way or another. Yet as of today, these devices—managed and secured at their user’s own discretion—remain a huge risk.
To truly address the problem, the United States and other governments will need to take a far more ambitious approach. First and foremost, SignalGate should serve as a wake-up call to rein in the mercenary surveillance industry. Governments should put in place much stronger controls for the sale and use of surveillance technology, including imposing transparency requirements and adding export regulations. Regulators must mandate stronger standards for the telecommunications sector writ large, as the United Kingdom has recently done, and hold telecommunications companies responsible for closing known loopholes in their signaling systems. Legislators must pass stricter privacy laws that prevent the unauthorized and uncontrolled sale of location data, advertising intelligence, and other personal data, and drastically improve oversight across law enforcement and intelligence agencies. Finally, law enforcement and government watchdogs must impose real penalties on those caught flouting government protocols for the handling of classified information.
The SignalGate episode is a terrible failure of operational security. But it is not a one-off mistake. It is a glimpse of a much more far-reaching crisis, one that threatens the United States and every other country that relies on the digital world in making its most important and sensitive decisions.
